I. Introduction
Unmanned aerial vehicles (UAVs) are widely used to map and monitor land with high resolution, low cost, and high speed. However, in hazy conditions, the collected images often suffer from low visibility, sharpness, and color saturation, which affects subsequent applications. Since most framing situations are almost vertical downward at a certain height (more than 500 m), hazy UAV images have different features than natural images.
The scene depth has little difference, which easily leads to loss of transmission details.
The transmission is quite small (mostly less than [1]) due to a longer scene depth. Most images easily result in energy over-reduction and color distortions during dehazing.